Meteoritics & Planetary Science 43, Nr 3, 479–496 (2008) AUTHOR’S Abstract available online at http://meteoritics.org PROOF

Geochemistry of Darwin glass and target rocks from Darwin crater, Tasmania, Australia

Kieren T. HOWARD

School of Earth Sciences, University of Tasmania, Hobart, Tasmania 7000, Australia Present address: Impacts and Astromaterials Research Centre, Department of Mineralogy, The Natural History Museum, Cromwell Road, London SW7 5BD, UK Corresponding author. E-mail: [email protected] (Received 1 March 2007; revision accepted 25 July 2007)

Abstract–Darwin glass formed about 800,000 years ago in western Tasmania, Australia. Target rocks at Darwin crater are quartzites and slates (Siluro-Devonian, Eldon Group). Analyses show 2 groups of glass, Average group 1 is composed of: SiO2 (85%), Al2O3 (7.3%), TiO2 (0.05%), FeO (2.2%), MgO (0.9%), and K2O (1.8%). Group 2 has lower average SiO2 (81.1%) and higher average Al2O3 (8.2%). Group 2 is enriched in FeO (+1.5%), MgO (+1.3%) and Ni, Co, and Cr. Average Ni (416 ppm), Co (31 ppm), and Cr (162 ppm) in group 2 are beyond the range of sedimentary rocks. Glass and target rocks have concordant REE patterns (La/Lu = 5.9–10; Eu/Eu* = 0.55–0.65) and overlapping trace element abundances. 87Sr/86Sr ratios for the glasses (0.80778–0.81605) fall in the range (0.76481–1.1212) defined by the rock samples. ε-Nd results range from −13.57 to −15.86. Nd model ages range from 1.2–1.9 Ga (CHUR) and the glasses (1.2–1.5 Ga) fall within the range defined by the target samples. The 87Sr/86Sr versus 87Rb/86Sr regression age (411 ± 42 Ma) and initial ratio (0.725 ± 0.016), and the initial 43Nd/144Nd ratio (0.51153 ± 0.00011) and regression age (451 ± 140 Ma) indicate that the glasses have an inherited isotopic signal from the target rocks at Darwin crater. Mixing models using target rock compositions successfully model the glass for all elements except FeO, MgO, Ni, Co, and Cr in group 2. Mixing models using terrestrial ultramafic rocks fail to match the glass compositions and these enrichments may be related to the projectile.

INTRODUCTION expected in small simple impact craters formed in sedimentary target rocks (Howard and Haines 2007). Glass Darwin glass is a siliceous impact glass found in a strewn distribution data are consistent with ejection from Darwin field covering more than 400 km2 in western Tasmania, crater and show: 1) the largest recovered fragments are found Australia. The glass also appears across the state in middens closest to the crater, 2) a decrease in the abundance glass and caves formerly visited by Tasmanian Aboriginals who are fragments away from the crater, and 3) an increase in the the traditional owners of the glass. Darwin glass has been proportion of splashform, relative to irregular or ropy, shapes repeatedly dated at ~800 ka (Gentner et al. 1973) and our away from the crater (Howard and Haines 2003; Howard most recent age estimate of 816 ± 7 ka was determined by Ar- 2004). Ar methods (Loh et al. 2002). Ford (1972; his Fig. 1) Impact glasses and tektites inherit the isotopic and identified a small (diameter = 1.2 km) circular depression geochemical signals of the source rocks melted during their near to the eastern edge of the strewn field in the Southwest formation. The aim of this paper is to test the compatibility of World Heritage Area (42°18.39′S, 145°39.41′E). This the suspected target rocks at Darwin crater as the materials pronounced topographic feature was named Darwin crater melted under impact conditions to form Darwin glass. Most and assumed to be the source of Darwin glass, despite a lack impact glasses and tektites have a composition similar to of conclusive evidence and a paucity of data on the glass or average upper continental crust and, as such, major- and trace- crater. The origin of Darwin glass has been investigated in a element composition alone is rarely enough to conclusively Ph.D. thesis (Howard 2004) and this paper stems from that relate a glass to a suspected target rock or to define an impact work. The geology of Darwin crater has been described in a origin. However, such investigations can effectively rule a detailed petrographic study and is consistent with that suspected target rock in or out of further consideration in

479 © The Meteoritical Society, 2008. Printed in USA. 480 K. T. Howard

Fig. 1. Darwin crater and Darwin glass strewn field. The grey shaded area defines the Darwin glass strewn field. studies aimed at defining the impact origin of a glass. In this quartzite the west. Rocks at the crater are correlates of the paper, the geochemical and Sr,Nd isotopic systematics in Eldon Group; a succession of low-grade metasedimentary Darwin glass are shown to be compatible with the suspected rocks consisting of quartzites and slates (Fig. 3). Gould (1866) target rocks at Darwin crater. first named these rocks the “Eldon Beds” and defined them as the rocks overlying the main limestone succession (Gordon STRATIGRAPHY Group), near the mouth of the Gordon and along the Eldon Rivers. Gill and Banks (1950) formally defined the Eldon Darwin crater is situated on the Queenstown 1:250,000 Group and the following formations are recognized: (top) Bell scale geological map sheet (Corbett and Brown 1975; Fig. 2), Shale; Florence Sandstone; Keel Quartzite; Amber Slate and which shows significant geological complexity across the area Crotty Quartzite (base, Fig. 3). The metamorphic grade of the of the strewn field. The densely rainforested valley where rocks is lower greenschist, with the development of aligned Darwin crater lies is a tributary of the Andrew River valley. The micas that define the foliation cleavage of the slate units. In Andrew River valley cuts into Ordovician limestone (Gordon some samples, minor (<2%) chlorite alteration is also observed. Group). Siliceous conglomerate (Denison Group), Two drill holes penetrate the center of the crater. The unconformably dipping off Proterozoic quartzite of the deepest hole penetrated to a depth of ~230 m. Drill cores Engineer Range, forms the east side of the valley, and Silurian intersected fine-grained lacustrine sediments (~60 m thick), Geochemistry of Darwin glass and target rocks from Darwin crater 481

Fig. 2. Geology of the Darwin glass strewn field. Based on Corbett and Brown (1975) and Corbett et al. (1993). overlying poorly sorted coarser crater-fill deposits. The pre- crater-fill samples is far greater than in rocks surrounding the lacustrine crater-fill comprises a complicated package of crater but diagnostic shock indicators (e.g., planar deformation angular polymict breccias of quartz and country rock with very features [PDFs] in quartz grains) have not been observed. The rare glass, monomict sandy breccias of angular quartz, and petrographic features of the suspected target rocks and crater-fill clasts of deformed slate. The degree of deformation evident in samples are described in detail in Howard and Haines (2007). 482 K. T. Howard

Fig. 3. Darwin crater geology. Based on field mapping in this study and by R. J. Ford; 1:25,000 scale aerial photographs; Corbett and Brown (1975) and Corbett et al. (1993).

METHODOLOGY and these were mounted in epoxy discs 25 mm in diameter. In addition, 26 glass samples <5 mm in size and showing Scanning Electron Microscopy (SEM) splashform (droplet, spheroid, and elongate) shapes (Fig. 4) were collected from across the strewn field, cleaned, and Small pieces of Darwin glass were carefully chipped mounted intact in 25 mm epoxy discs. These “mini-glass” from individual larger fragments and placed in an ultrasonic grains were polished carefully to expose a fresh interior bath in distilled water for cleaning. The studied glass surface for analysis. This is the first time such splashform fragments were collected from across the strewn field. The mini-glass samples of Darwin glass have been reported or detached pieces’ sizes ranged from 2 to 10 mm in diameter analyzed. Geochemistry of Darwin glass and target rocks from Darwin crater 483

612, and the standard BCR-2 was analyzed as an unknown. During the analyses, two analyses of both NIST 612 and BCR-2 were made before ten analyses of Darwin glass, followed by two more analyses each of NIST 612 and BCR-2. As an internal standard the measured intensity of 49Ti during each analysis was normalized to the TiO2 content of each glass determined previously by SEM. Throughout the course of the study, detection limits were between 0.1 and 0.001 ppm. Spectra for data presented here are flat and smooth with little deviation in counts per second throughout the 60 s duration of each analysis. This is generally the case in analyses of tektites and impact glasses that are well suited to the LA-ICPMS technique, especially for determination of refractory elements (rare earth elements [REE], Sr).

Suspected Target Rock Sample Selection and Preparation

Fig. 4. Splashform Darwin mini-glasses. Scale bar = 1 cm. In total, 33 samples that encompass the range of Eldon Group rocks found around the crater and throughout the drill Major elements were determined using a CAMECA cores were selected for analysis. Most of the analyzed surface SX50 scanning electron microscope, in WDS mode, at the samples were collected along a W-E traverse from the access Central Science Laboratory (CSL), University of Tasmania. track, to the eastern edge of the valley hosting the buried The regulated electron beam current was operated at 25 nA at crater. These samples were collected in situ within the dense an accelerating voltage of 15 kV. A nominal incident beam forest and not from bulldozer scrapings at the tracks edge or size of 8 µm diameter was used in order to minimize alkali sampled regionally. The Crotty Quartzite formation is migration and consequent elevation of Si and Al counts. Eight mappable from air photos but sampling the lithology very major elements (Si, Al, Fe, Mg, Ti, K, Ca, Na) were analyzed near to the crater is next to impossible owing to a lack of using the mineral standards and calibrations provided by Dr. outcrop and the extreme vegetation at the eastern edge of the D. A. Steele. Detection limits range between 0.05 and 0.1 wt%. crater. For detailed chemical analyses (e.g., Sm,Nd isotopes), Sodium was analyzed first to reduce the effect of volatilization only stratigraphically constrained samples of Keel Quartzite on the analysis. 2 spots were analyzed on each of the glass and Amber Slate from the craters edge were studied. This chips (n = 216). For the <5 mm mini-glasses, three spots were stratigraphic constraint was lacking in previous studies— analyzed on each grain (n = 78). This results in a total of 294 Meisel et al. (1990) are likely to have analyzed samples from analyses and significantly expands the published analyses of the Florence Sandstone and the Bell Shale exposed by Darwin glass, the next largest study being Meisel et al. bulldozer during access track constructions. The position of (1990), who analyzed 18 samples. Howard (2004) also analyzed core samples is depicted in Fig. 5. contains more than 1000 other glass analyses from grid surveys Outcropping suspected target rock samples were on individual glass fragments. carefully selected to avoid analyses of highly weathered samples. These hand-sized specimens were split into ~10 mm Laser Ablation Inductively Coupled Plasma chips and the weathered material was discarded. The chips Mass Spectrometry (LA-ICPMS) were further crushed using a hydraulic crusher and cleaned of dust with an air hose. In analyses that exclude Co, around The same 25 mm epoxy discs containing Darwin glass 40 g of the crushed material was ground in the tungsten fragments that were analyzed by SEM were analyzed for trace carbide ring mill. For most analyses that include Co, the element compositions using LA-ICPMS. The School of Earth crushed material was further pulverized in an agate mortar Sciences at the University of Tasmania uses an Merchantek and pestle before being further crushed to fine powder in an 266 nm laser operated at 10 HZ and 3–6 J/cm2 per pulse and agate ring mill. the Agilent HP 4500 ICP-MS. 23 elements were determined during each analysis and these were: Cs, Rb, U, Th, Ba, La, X-Ray Fluorescence (XRF) Analyses Ce, Nb, Pr, Sr, Nd, Zr, Sm, Eu, Gd, Ho, Yb, Y, Lu, Sc, Cr, Co, and Ni. A nominal beam size of 120 µm was used throughout Fusion discs of the powdered target rock samples were all analyses. Again two spots were analyzed on each glass prepared for determining the major elements (SiO2, Al2O3, chip for a total of 216 analyses. Each of the 26 mini-glasses TiO2, MnO, MgO, FeO, K2O, CaO, Na2O, P2O5) by X-ray were analyzed at four points each. Relative element fluorescence (XRF). Pressed pills were used for XRF analysis sensitivities were calibrated against the glass standard NIST of the following trace elements: Cr, Co, Ni, Cu, Zn, Ga, Nb, 484 K. T. Howard

Fig. 5. Darwin crater drill core stratigraphy. The location of samples selected for geochemical analyses are indicated.

Rb, Sr, Ba, Zr, and Y. These analyses were conducted on an analyses using ICP-MS. This technique was chosen primarily automated Phillips PW 1480 XRF spectrometer at the School to determine the REE geochemistry of the suspected target of Earth Sciences, University of Tasmania. Mr. Phillip rocks, but trace elements determined by XRF were also Robinson provided calibrations. analyzed by ICPMS and it is these that are reported here. The following trace elements were determined: Sc, Cr, Co, Ni, Ga, Solution ICPMS Rb, Sr, Y, Nb, Sb, Cs, La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Yb, Lu, Hf, Ta, Pb, Th, and U. The abundance of the trace Portions of samples (KH 1,2,4,6,7,8,12,15,17,18,19) that elements in these samples was calibrated against the rock were analyzed by XRF were dissolved using HF and HNO3 standard BHVO-1. Instrument drift was monitored and acids and brought to a 1:1000 dilution in 2% HNO3 to allow corrected using La and Ti as internal standards. Geochemistry of Darwin glass and target rocks from Darwin crater 485

Table 1. Major-element composition of Darwin glass. Site SiO2 Al2O3 TiO2 FeO MgO CaO K2ONa2O n Macro-glass 76.47–93.85 3.14–11.45 0.22–0.8 0.84–5.87 0.24–4 0.02–0.25 0.75–2.71 0–0.21 Average 84.6 7.52 0.58 2.55 1.13 0.06 1.87 0.1 216 Mini-glass 75.24–93.18 1.98–10.68 0.13–0.7 0.74–6.2 0.19–3.67 0–0.22 0.61–2.4 0.02–0.19 Average 84.41 7.37 0.56 2.87 1.56 0.09 1.71 0.08 78 Meisel et al. 84–89.3 6.75–8.20 0.52–0.62 1.08–3.78 0.61–1.13 0.03–0.18 1.51–2.93 0.02–0.06 18 (1990) Taylor and 84.1–87.1 5.8–7.44 0.56–0.62 1.44–2.98 0.66–1.36 0.05–0.19 1.66–1.98 0.03–0.07 10 Solomon (1962) Analyses by SEM; detection limits are between 0.05–0.1 wt%.

Isotopic Analyses and FeO (+0.27%) and depleted in K2O (−0.2%). This variation is insufficient to define the mini-glasses as geochemically Rb-Sr and Sm-Nd isotope systematics were studied in 12 distinct from the macro glasses. On the basis of major- samples: three slates (Amber Slate); three quartzites (Keel element geochemistry, all of the analyzed samples are Quartzite); three crater-fill samples (Fig. 5) and three pieces of considered to be parts of the same population. These mean glass. The analyzed slates and quartzites samples were results are similar to previously published analyses of Darwin collected proximal to the crater; isotope abundances were not glass (e.g., Taylor and Solomon 1962; Meisel et al. 1990). determined for Crotty quartzite owing to the lack of such a However, average Darwin glass in this study is depleted in stratigraphically constrained sample from close to the crater. SiO2 and enriched in MgO relative to previous studies. This Rb, Sr, Sm, and Nd concentrations and isotope compositions study extends the ranges in composition of Darwin glass for were determined by Dr. Karin Barovich at the University of all analyzed major elements. Adelaide. Sr and Nd isotopic compositions were measured in static mode on a Finnigan MAT 262 thermal ionization mass Trace Elements in Darwin Glass spectrometer. The average 87Sr/86Sr ratio for standard NIST SRM987 during the course of the study was 0.710264 ± 28 (2σ) A summary of the mean and range of trace element on 12 runs. The procedural blanks for Sr are better than 1 ng. abundances in macro and mini Darwin glasses is presented in Whole rock powders were dissolved and split before spiking Table 2. All of the glasses show affinity with upper crustal with 84Sr and 85Rb. Samples for the Sm-Nd analyses were sediments and are relatively enriched in light REE (LREE) spiked with a mixed 149Sm-150Nd spike. The value for standard (La/Lu = 5.9–10) with pronounced negative Eu anomalies Lajolla gave a 143Nd/144Nd ratio of 0.511848 ± 8 (2σ). The (Eu/Eu* = 0.55–0.65). Macro and mini Darwin glasses have procedural blanks for Nd are less than 300 pg and for Sm less overlapping compositional ranges and very close average than 150 pg. compositions for all elements except Cr, Co, and Ni that are enriched in the mini-glasses. With the exception of Ni RESULTS (enriched by >30%), the results are within 20% of previously published analyses (e.g., Meisel et al. 1990). These data Major Elements in Darwin Glass extend the previously reported compositional ranges for the trace elements in Darwin glass. The variation in trace element The ranges in composition for the major elements in the compositions in Darwin glass is strongly affected by SiO2 analyzed macro Darwin glasses are: SiO2 (76.5–93.9%), closure. This artefact produces a statistically significant Al2O3 (3.1–11.4%), TiO2 (0.2–0.8%), FeO (0.8–5.9%), MgO negative correlation between SiO2 and all trace elements. (0.25–4.0%), K2O (0.7–2.7%), CaO (<0.01–0.3%), and Na2O A cluster analysis performed on the following major- and (<.01–0.2%). For the analyzed mini-glasses, the ranges in trace-element data: Na2O, MgO, Al2O3, SiO2, K2O, CaO, TiO2, major-element composition are: SiO2 (75.24–93.18%), Al2O3 FeO, Sc, Cr, Co, Ni, Rb, Sr, Zr, and Ba resulted in a dendrogram (1.98–10.68%), TiO2 (0.13–0.7%), FeO (0.74–6.8%), MgO with two main compositional groups (Table 3). Group 1 (0.19–3.67%), K2O (0.61–2.4%), CaO (<0.01–0.22%), and includes 80% of the sample and is characterized by large Na2O (<.01–0.19%). A summary of the mean and range in the variations in major-element compositions. In hand samples and major-element composition of Darwin glass is presented in thin sections, glasses belonging to group 1 are predominantly Table 1. Macro and mini Darwin glasses have overlapping ranges white to dark green in colour. The range in major-element in major-element compositions and very similar average composition in group 1 glass is: SiO2 (80.62–93.9%), Al2O3 compositions, however, relative to average macro Darwin glass, (3.14–10.6%), TiO2 (0.2–0.76%), FeO (0.8–4.23%), MgO the average mini-glass is slightly enriched in MgO (+0.46%) (0.25–2.31%), and K2O (0.7–2.7%). The average composition 486 K. T. Howard

Table 2. Trace element composition of Darwin glass. Meisel et al. Taylor and Solomon Macro-glasses Mini-glasses (1990) (1962) Range Average Range Average Range Range (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) Cs 1.5–5.8 3.8 1.6–6.7 3.3 2.5–4.5 3.2–4.3 Rb 33.3–109.3 75.3 35.5–118.1 67.7 71–137 61–110 U 0.6–4 1.9 0.4–3.6 1.5 1.5–5.4 – Th 4.5–22.8 14 6.6–18.8 13.9 12–19 – Ba 116.7–457.2 304.9 166.8–384.7 293.5 182–450 290–360 La 48.9–11.3 36.2 17.1–45.5 35.1 35–46.5 – Ce 26.9–110.3 79.4 41.2–105.2 78.3 70–97.8 – Nb 4.6–16 11.6 6.0–14.6 11.2 – – Pr 2.7–12.5 8.7 4.4–11.3 8.5 – – Sr 4.9–27.8 15.6 7.9–24.3 14.8 – 13–16 Nd 10.7–48.2 33.4 16.1–43.4 32.2 29–42 – Zr 54.1–50.9 433.2 180.6–919.6 416.7 254–547 220–490 Sm 2.1–10.7 6.9 3.4–8.6 6.7 6.6–9 – Eu 0.4–1.9 1.3 0.7–1.8 1.3 0.9–1.5 – Gd 1.9–11.1 6.6 3.3–8.2 6.2 – – Ho 0.3–2.0 1.2 0.6–1.7 1.2 – – Yb 0.8–4.9 3.3 1.3–4.4 3.1 2.7–4.5 – Y 7.4–53.4 34.1 14.2–45.4 31.9 – 18–44 Lu 0.1–0.8 0.5 0.1–0.6 0.5 0.1–0.8 – Sc 3.5–10.8 7.3 3.9–9.8 6.8 6–8.2 2.5–5.2 Cr 19.5–505.2 89.9 39.4–371.7 120.2 48–522 69–205 Co 0.3–56.7 12.7 4.2–59.7 20.4 4.6–39 <3–27 Ni 3.0–917.7 161.3 34.7–841.8 246.2 30–536 82–205 Analyses by LA-ICPMS; detection limits between 0.1 and 0.0001 ppm. of group 1 glass is close to that of bulk average Darwin glass. Amber Slate (75.41%). For most major elements, the The second population identified in the cluster analysis is deformed rocks from the drill core show obvious affinity characterized by a narrower range in and lower average with the Amber Slate and Keel Quartzite (KH17) cropping abundance of SiO2 (76.5 to 84.1%, with an average of 81.2%). out around the crater. In some samples the effects of In hand samples and thin sections, group 2 glass is almost weathering on the crater-fill are obvious. For example, in always black. The average MgO (2.2%) and FeO (3.8%) sample KH19 (taken from a highly fractured and brecciated compositions of group 2 are significantly higher than those of zone filled by the oxyhydroxide goethite), the FeO content group 1 glass or average Darwin glass, and Al2O3 is also reaches 33.5% at just 37% SiO2. The polymict, allogenic slightly enriched. The average Cr (162 ppm), Co (31 ppm), breccia samples (crater-fill facies A) also show strong and Ni (416 ppm) content of group 2 glass is also geochemical affinity with outcropping surface rocks and the significantly enriched relative to group 1 glass or average deformed shales intersected in the drill cores, except for up to Darwin glass, with the remaining trace elements being of very 5% S in samples KH25 and KH26 from the top of this facies. similar abundance in both groups. On the basis of 18 samples, This elevated S content is consistent with petrographic Meisel et al. (1990) identified 3 chemically distinct glass observations and XRD data that indicate the secondary groups and importantly one sub-population enriched in Mg and mineral Rosenite (Howard and Haines 2007). the transition metals was identified. Taylor and Solomon (1962) also reported anomalous enrichments in Ni, Co, and Cr in some Trace Elements in Suspected Target Rocks and Crater-Fill Darwin glasses. Samples from Darwin Crater

Major Elements in Suspected Target Rocks and Crater- The range and average abundance of trace elements in Fill Samples from Darwin Crater surface and crater-fill samples from Darwin crater are presented in Table 5. For most elements, the slates show the most limited Analyses of surface Eldon Group rocks and crater-fill compositional range. In all analyses the REE—especially the samples from the drill cores are presented in Table 4. The LREE—show the least variation and range from between <2% average SiO2 content is highest in the surface samples of (e.g., Sm in Amber Slate) and 60% (e.g., Gd in Keel Quartzite) Crotty (94.6%) and Keel (90.09%) Quartzite, followed by of mean values. The transition metals Ni and Co and the alkaline Geochemistry of Darwin glass and target rocks from Darwin crater 487 0.9 116.7–457.1 6–553.1 210.7–427.3 neity. Average group 1 glass is close Average neity. FeO and MgO contents. Group 2 is also 74.75 71.1–76.91 37.92 7.03 ite) breccia) (Polymict abundance with significantly higher 2 abundance and a large degree of heteroge abundance and a large 2 6 0.0–33.9 3.0–492.8 33.2–109.2 4.9–27.8 54.1–75 formed slate and quartz Crater-fill facies B and C Crater-fill facies A Crater-fill 6–260.4 19.4–56.5 117.4–917.7 56.6–109.2 10.9–20.4 286. 0 0.09 <0.03–0.14 0.09 0.06–0.14 0.09 <0.03 wide range in SiO 365917 0.68 6.1815 0.9724 0.53–0.84 0.277 3.2 1.6–10.75 0.32–1.53 0.6 0.19 0.05–1 4.03 1.11–4.69 0.96 0.037–0.46 0.36–0.78 2.07–6.88 0.66 0.52–1.28 0.41 2.79 33.52 0.63 0.12 0.06–2.44 0.08 1.43–3.79 1 1.04 0.55 0.01–0.3 2.13 1.75 49.94 0.3 0.06 94 11.32 4.43–9.26 9.55 4.95–12.6 7.46 1.17 range and a lower average SiO FeO Sc Cr Co Ni Rb Sr Zr Ba characterized by a 2 94.60 93.24–97.6 70.54 53.38–72.81 TiO glass compositions. .0–0.1 0.2–0.7 0.8–4.23 3.4–10.1 19.5–204. CaO 7 0.06–0.2 0.5–0.80 1.8–5.8 6.3–10.7 67. has a more limited compositional O 2 K target rocks and crater-fill facies. rocks and crater-fill target 2 sification of Darwin SiO gram with 2 compositional groups. Group 1 is 3 to group 1 and bulk average glass. O 2 Darwin glass. Group 2 glass Al MgO Keel Quartzite Amber Slate Quartzite Crotty (De O 2 5.430.490.420.35 4.13–8.440.01 0.47–0.560.08 0.31–0.621.82 11.83 0.27–0.510.03 0.710.07 3.28 0.01–0.296848511 10.29–14.6 0.01 1.44–2.81 0.01–0.05 0.56–0.86 0.08 2.82 0.01–0.3 1.88–4.57 3.47 0.01–1.58 <0.01 0.24 0.20 0.35 0.14 0.01–0.32 1.11–3. 0.15 2.92–4.04 <0.01–0.02 0.14–0. 0.11–0.42 0.07 0.22–0. <0.01 0.99 0.1–0.35 0.09–0. 0.02 0.01–0. 0.70 0.38–1. 0.0015–0. 0.01–0. 0.19 <0.01–1.01 0.03 <0.01–0.21 0.2 0.08 Na 0.1 1.1 7.5 84.6 1.9 0.1 0.6 2.6 7.3 89.9 12.7 161.3 75.3 15.6 433.2 304.9 Average Range Average Range Average Range Average Range Average Range KH19 KH29 90.09 85.63–92.63 75.41 70.14–79.05 3 5 2 2 O O 2 2 O 2 Average Average group 1 0.05Range Average 0.9 0.0–0.2group 2 0.2–2.3 0.1 3.1–10.6Range 7.3 80.62–93.9Bulk 0.0–0.2 0.75–2.6 2.2average 0 1.1–4.0 85.2 Darwin 6.4–11.5glass 8.2 76.4–84.5 1.8 1.4–2. 81.6 0.05 2 0.05 2.2 0.1 7.2 0.6 74.5 3.8 8.1 8.7 162.7 108.4 31.6 74.6 416.6 15.8 78.1 438.9 14.3 304.3 439 294 Group The cluster analysis results in a dendro SiO Al TiO FeO MgO MnO CaO K2O Na P n Determined by XRF. in composition to bulk average enriched in Ni, Co, and Cr relative Table 3. Cluster analysis group clas Table composition of 4. Major-element Table 488 K. T. Howard

Fig. 6. Chondrite normalized REE composition of Darwin glass, target rocks, and crater-fill samples. Chondrite values from Sun and McDonough (1989). earth Sr show the greatest compositional ranges of the • The glasses are very slightly depleted in LREE relative determined elements. This large variation in analytical results— to average Amber Slate and very slightly enriched in La particularly for the transition metals—relates to the dilution and Ce relative to average Keel Quartzite. The effect of quartz and the subsequent very low abundance of these remaining LREE abundances are effectively identical in elements in the most quartz-rich samples. The ranges in the both Darwin glass and Keel Quartzite. For the other abundance of all trace elements in the crater-fill samples fall HREEs and Y, the glasses and suspected target rocks largely within the total ranges for the analyzed Eldon Group have overlapping abundances similarly shaped patterns rocks from around the crater. The obvious exceptions being (Fig. 6). enrichments in heavy REE (HREE) plus Y, transition metals Ni, • For the alkali metals, the alkali earth Ba, and actinide Co, Cr and the sulfide metals Zn, As, Cu, and Pb in crater-fill elements, the glass compositions overlap the target rocks facies C. This enrichment is interpreted to be related to and are closest to average Keel Quartzite (Fig. 7). weathering processes and is most pronounced in those analyzed • The average Zr abundances in the glass falls within the samples with the most abundant Fe-oxyhydroxides such as range in Zr abundances of the suspected target rocks KH19 that has up to 150 ppm Ni, 190 ppm Cr, and 720 ppm Y! (Fig. 7). • Average Rb/Sr in group 1 (4.8) and 2 (5.2) glasses falls GLASS VERSUS TARGET within the compositional range defined by average Crotty Quartzite (3.55), Keel Quartzite (7.17) and Amber Major and Trace Element Data Slate (9.98). The strongly depleted concentration of Sr and high Rb/Sr ratios in both the glass and suspected A comparison of the composition of Darwin glass and target rocks, relative to average crustal rocks, is the key target rocks from Darwin crater shows: feature of the non-transition metal trace element • Major elements in group 1 Darwin glass are closest to chemistry that links the glass to these target rocks. average Keel Quartzite (Sk) but are dramatically enriched • The upper limits in the abundance of Ni, Co, and Cr in in FeO and also enriched in MgO, P2O5, and SiO2 relative group 2 Darwin glass samples are more than 20, 6, and 2 to the quartzite. In group 1 glass, all major elements other times the average suspected target rock values, than SiO2 are depleted compared to average Amber Slate respectively. (Sa). Group 1 glass is enriched in all elements other than SiO2, MgO, CaO and P2O5 relative to average Crotty Isotope Data Quartzite (Sc). Group 2 glass is most similar in major- element composition to Amber Slate. Results of isotopic analyses (Table 6) show: Geochemistry of Darwin glass and target rocks from Darwin crater 489 Sr Error 86 Sr/ 87 Sr Error 86 .7–63.5 859.83 23.30 .6–43.2 153.93 5.70 Darwin glass. Rb/ 12.1–37.2 16.81 426.60 87 14 236.1–345 265.10 33.40 75.72 55.4–85.5 190.30 8.20 8–627 428.48 248.3–578.7 371.80 56.50 r target rocks, crater-fill samples and rocks, crater-fill r target formed slate and quartzite) (Polymict breccia) Crater-fill facies B and CCrater-fill facies A Crater-fill KH19 KH29 10.77 1–19.1 14.66 5.4–20.7 15.31 <2 52.17 33.19–85.8 41.40 19 12.97 6.65–28.82 24.02 14.83 1505.72 80.13 15.80 14.66 0.59 0.81 0.00 15.11 1478.66 82.90 17.60 13.41 0.54 0.82 0.00 14.54 1504.93 82.60 17.20 13.66 0.55 0.81 0.00 14.83 1533.57 74.90 12.60 16.90 0.68 0.81 0.00 14.03 1572.43 105.90 14.32 15.39 0.62 0.81 0.00 14.16 1386.29 171.80 7.49 n.d n.d n.d n.d 13.57 1744.19 49.70 6.65 21.34 0.85 0.85 0.00 14.35 1586.81 96.20 28.82 9.45 0.38 0.76 0.00 15.70 1335.30 178.17 15.85 25.00 1.00 0.88 0.00 15.59 1239.93 197.70 9.00 n.d n.d n.d n.d 15.86 1402.00 157.10 18.37 24.47 0.98 0.87 0.00 15.64 1363.96 179.70 20.17 25.52 1.02 0.88 0.00 14.82 1545.47 86.03 12.78 31.73 1.27 0.91 0.00 15.17 1430.54 63.00 11.05 16.24 0.65 0.82 0.00 13.58 1932.72 84.60 22.56 10.64 0.43 0.78 0.00 15.70 1273.14 110.50 4.74 68.31 2.73 1.12 0.00 − − − − − − − − − − − − − − − − 6 147.80 49.7–197.3 123.70 68.9–158.7 83.40 13.80 7 42.76 7.6–93.5 30.54 15 1 15.65 10.55–20.2 13.44 8.2–17.8 10.74 1.50 –305.5 319.99 254–440.8 290. ange Range Average Range Average Nd Error e—Nd tCHUR (ma)(ppm) Rb (ppm) Sr 144 Nd/ 143 83 76.34–87.1 90.64 44.7–119.9 Nd Error 7.3 124.52 108.9–143.97 459.43.72 181. 40.47 33.95–50.25 106.44 144 7–0.535–12.65 0.60 9.11 0.44–0.88 4.66–13.42 13.71 94.75 8–1.395–6.254–1.25 1.82 8.65 1.57 1.62–2.19 7.46–10.84 1.24–2.17 11.68 69.06 24.08 4–49.441–966–11.12 41.66 87.66 33.91–51.38 10.40 73.22–107.21 8.57–12.9 87.42 161.56 22.32 1–12.354–20.54 9.50 17.34 3.98–13.92 14.61–21.57 9.65 21.98 Sm/ 147 ations, isotope ratios, and Nd model ages fo (4%). σ target rocks and crater-fill facies. rocks and crater-fill target t. Errors are 2 in Excel using Isoplo Keel Quartzite Slate Amber Crotty Quartzite (De Average 6.95 35.06 0.12 0.00 0.51 0.00 Average 8.18 39.46 0.13 0.00 0.51 0.00 Average 6.66 38.5 0.1 0.00 0.51 0.00 Average 5.93 29.66 0.12 0.00 0.51 0.00 Average Range Average Range Average R Sample Lithology Sm (ppm) Nd (ppm) KH24 Black glass 6.74 34.82 0.12 0.00 0.51 0.00 KH23 Dark green glass 7.12 35.45 0.12 0.00 0.51 0.00 KH22 Light green glass 7 34.91 0.12 0.00 0.51 0.00 KH18 Crater-fill 9.32 48.16 0.12 0.00 0.51 0.00 KH17 Crater-fill 7.63 33.92 0.14 0.00 0.51 0.00 KH15 Crater-fill 7.58 36.29 0.13 0.00 0.51 0.00 KH8 Amber Slate (Sa) 6.61 40.53 0.1 0.00 0.51 0.00 KH7 Amber Slate (Sa) 6.23 34.73 0.11 0.00 0.51 0.00 KH6 Amber Slate (Sa) 7.13 40.24 0.11 0.00 0.51 0.00 KH4 Keel Quartzite (Sk) 4.86 25.8 0.11 0.00 0.51 0.00 CoNi 1.43 7.42 1–2.4 0.5–12.3 21.53 3.65 9.2–29.1 1–6.2 4.35 5.15 0.5–9. 2.2–8.1 LuScCr 0.42 6.09 54.52n6Determined by XRF and solution ICP-MS. 0.34–0.58 4.33–6.97 37.3–57.1 0.50 11.20 85.39 0.4 8 9.9 72–100 80. 4 8 5 1 1 KH2 Keel Quartzite (Sk) 8.35 35.57 0.14 0.00 0.51 0.00 SmEuGd 6.31Ho 1.14Yb 5.64Y 1.07 4.96–8.67 2.82 24.19 0.83–1.64 4.01–8.42 0.87–1.48 7.31 2.26–3.91 1.29 5.96 6.81–7.63 9–39.67 1.18 1.1 3.34 5.6 31.39 1.1 3.18–3.6 26.2–35.8 15.63 9–21.2 9.09 7.87–11.16 4.06 3.31–5.86 36.26 81.88 LaCeNb 31.33Pr 65.94Sr 8.83 7.85 29.77–34.19 8.76 61.59–74.09 46.45 4.1–12.5 90.91 7.19–8.96 43.8 1.2–22.56 85.5 16.36 10.64 16.15 13.8–19.83 9.9 5.23 9–22.9 10.15 4–8. 1.2–20.6 Cs 4.27RbUTh 62.85Ba 4.12–5.51 2.17 10.58 232.52 14.6–110.5 9.69 1.75–2.73 161.24 7.89–13.37 108.9–350.5Nd 7.7 562.01 17.85 128.2–197.7 29.90 3.33 36.13 503.7–70 15.5 2.91–3.83 26.61–35.35 14.6–45. 38.22 35.66–39 3.24 1.99–4.11 7.94 Zr 309.82 112.9–600.3 336.79 222.6–404 171.24 112.9 Regression calculated KH1 Keel Quartzite (Sk) 4.59 27.63 0.1 0.00 0.51 0.00 Table 5. Trace element composition of 5. Trace Table 6. Rb-Sr and Sm-Nd elemental concentr Table 490 K. T. Howard

• 87Sr/86Sr ratios for the glasses fall within the very large consistent with the interpretation that Darwin glass can be range (0.76481–1.1212) defined by all suspected target formed from impact melting of the suspected target rocks. rocks and crater fill samples. Meisel et al. (1990) could also not reconcile the transition metal • The glass, suspected target rock, and crater-fill samples composition of some Darwin glass samples with the lower have homogeneous ε-Nd results ranging from −13.57– abundances measured in their studies of suspected target rocks. 15.86 with the glasses falling close to the middle of this However, Meisel et al. (1990) could also not reconcile the compositional range (−14.54–15.11; Fig. 8). Nd model lower abundances of Na, K, Rb and Cs in the glass relative (CHUR) ages for all analyzed samples range from 1.2– to their suspected target rocks and suggested selective 1.9 Ga using a reference chondritic mantle reservoir volatilization of these elements during impact. Based on data (CHUR) and the glasses (1.2–1.5 Ga) fall within the range presented in this study, the evidence for volatilisation is very defined by the suspected target and crater-fill samples. limited. P2O5 is the only element that is depleted in average • On isochrons the glass, suspected target rock, and crater- glasses relative to the target rocks. P is typically considered to fill samples define linear trends with glasses falling near to be a refractory element so a volatility control on its abundance the center of the data array in plots of both of 143Nd/144Nd in the glass would be surprising. Rather, the apparent loss of 147 144 87 86 87 86 versus Sm/ Nd and Sr/ Sr versus Rb/ Sr P2O5 in average glass relative to the target rocks is likely to be (Figs. 9a and 9b). A Rb-Sr regression line through all of the an artefact that reflects the wide variability in sedimentary rock analyzed samples yields an age of 411 ± 42 Ma (2σ) and an compositions, and uncertainties in determining the indigenous 87 86 ± σ initial Sr/ Sr value of 0.725 0.016 (2 ). In the Sm-Nd P2O5 contributions. Here, the average P2O5 abundances over system, the intial 143Nd/144Nd ratio is 0.51153 ± 0.00011 the bulk rock melted to form the glass may have been (2σ) and the regression yields an age of 451 ± 140 Ma (2σ). significantly less than the average values determined for the These age estimates overlap and are consistent with each target rocks. The natural variability in sedimentary rock other within the error limits of these data. compositions may also explain the lower abundances of Na, K, Rb and Cs in the glass relative to the target rocks found by MIXING MODELS Meisel et al. (1990). It is also seems that the suspected target rocks in the Meisel et al. (1990) study may have been sampled To further test the compatibility of the analyzed rocks as from the Florence Sandstone and Bell Shale some 2 km from parent materials of Darwin glass, a least squares regression the crater during access track constructions. model was created in Excel. The preferred model is a 3- component mixture of average suspected target rock Sm Nd Isotopes compositions (Table 7). Relative to the bulk average abundances the worst model fits are (in order) Ni, Co, MgO, The Sr and Nd isotopic composition and regression ages Cr, and FeO. For the remaining elements the model produces are an inherited signal in the glass and do not reflect the glass a good match using an average target rock mixture of 43% formation age or time of impact that is placed at 816 ± 7 ka Amber Slate (Sa), 24% Keel Quartzite (Sk) and 30% Crotty (Loh et al. 2002). Being relatively “young,” the glass has not Quartzite (Sc) (Table 7). been significantly altered by water and shows no evidence of The average of all individual model results for group 1 devitrification. Analysed samples all contain primary textural and 2 glasses have been separated and are compiled in Table 7. features such as flow banding, “hot break surfaces” and As is expected for group 1 glasses, these results are almost surface vesicles. As such, surface flows or groundwater springs identical to those for the model of bulk composition using an are not considered to have imprinted a secondary isotopic average suspected target rock mixture of 43% Sa, 27% Sk and signal on the glass. In the case of the Rb-Sr system, the initial 30% Sc. In group 2, the average model result was 66% Sa, 4% Sr ratio (0.725 ± 0.016 [2σ]) resulting from the regression is Sk, and 30% Sc. This mixture results in increased residual close to, or within error of, the estimated value for Silurian errors for (in order) Ni, Co, Rb, MgO Cr, and FeO. The seawater (0.70875) and is also close to the value for modern concentrations of the remaining elements are successfully seawater. The regression age might be considered to reflect modelled. When soil (e.g., KH15) or weathered samples the depositional age of the sediments, as 411 Ma is very close (e.g., KH19) more enriched in transition metals are added to to the Siluro-Devonian boundary. However, given the the mixing models the residual errors in Ni, Co, Cr, MgO, and analytical errors, the Rb-Sr regression age is also within error FeO are not improved (Table 7), even if soil or weathered of the age of the Tabberabberan Orogeny that commenced at material is allowed to contribute at an unrealistically large around 395 Ma (Williams 1989) and this may have reset the level to the mix. system. The isotopic composition and regression ages in the samples may reflect a combination of a depositional age signal DISCUSSION and the effects of the orogeny. However, this interpretation assumes that the Rb-Sr and Sm-Nd fractionation, as measured in With the exception of the transition metals in some group the impact glasses, reflects processes that occurred during 2 glass, all of the geochemical and isotopic data presented are deposition or later diagenesis and deformation of the target Geochemistry of Darwin glass and target rocks from Darwin crater 491 tage contributions Model result nt that the percen (Proportion of each component mixed) Sa Sk Sc Soil KH19 model was run subject to the constrai sis with target rock mixtures. The rock mixtures. with target sis r every glass analy Cr Co Ni Y Rb Zr Sr Ba O 2 K .3 3.5 85.9 3.7 21.5 31.4 161.2 336.8 16.2 562.0 compositional match fo FeO MgO 3 O 2 Al 2 TiO 2 uares model was used to find the best 1.01.2 0.1 10.6 0.61.0 7.9 0.9 0.1 36.3 0.4 39.7 0.50.9 0.5 25.9 0.81.1 0.0 33.8 37.6 0.3 5.8 0.9 72.8 9.3 0.4 13.5 148.3 1.5 91.94.9 65.3 26.0 26.9 9.2 1.25.8 0.1 164.1 34.4 5.9 25.9 12.2 42.1 0.7 0.6 101.2 156.3 36.1 109.9 7.9 76.2 22.2 0.7 3.52.4 9.7 336.3 25.6 27.5 357.1 17.6 22.8 53.8 0.52.8 0.1 393.0 46.2 165.0 17.6 46.8% 10.7 0.3 23.5% 13.4 0.6 6.3 3.6 65.4 29.7% 8.4 31.4 43.3 34.9 0.6 23.70.3 42.0 7.1 25.0 108.2 55.8 15.4 43.3% 0.50.3 0.1 27.0% 43.9 2.5 145.7 90.3 50.0 29.6% 12.0 0.4 20.6 0.6 118.9 21.6 7.4 65.7% 8.5 30.6 34.1 0.6 18.1 13.7 4.1% 9.4 22.8 74.7 156.2 30.2% 36.1 0.5 44.1 21.9 146.4 3.4 91.9 0.4 19.5 22.4 11.9 7.6 36.4 31.1 34.6 23.7% 23.7 17.9 9.5 38.1% 75.2 146.5 34.0 17.3% 20.9% 25.1 146.3 3.9 91.8 28.5 12.6 9.7 38.3 35.4% 22.8 17.2 40.7% 149.9 34.8 22.9% 26.3 4.1 12.2 37.1 2.2% 34.9% 40.3% 22.2% 0.2% 2.4% SiO 85.5 0.6 6.8 2.3 0.8 1.7 69.4 8.2 110.1 36.0 66.2 457.4 16.4 238.0 Excel, the lease sq can not be negative. from each target rock from each target Using the Solver function in Model component compositions Amber Slate (Sa)Average 75.4 0.7 11.8 3.3 1 Average Keel Quartzite (Sk)Average Crotty Quartzite (Sc)Average SoilKH19Bulk average glass 90.1 94.6 group 1 glassAverage group 2 glassAverage 0.5 0.2 5.4 2.8 0.4 0.3 85.2 84.6 0.4 81.6 0.1 0.6 0.6 53.5 1.8 0.6 37.9 1.0 7.3 7.5 0.5 55.8 8.2 0.4 80.0 2.2 2.5 9.2 1.4 3.8 7.5 5.2 0.9 1.1 10.2 33.5 7.4 2.2 4.4 1.9 1.9 0.7 24.3 0.6 15.6 2.0 74.5 62.9 89.7 1.9 36.1 160.7 2.1 309.8 12.6 171.2 8.8 30.4 190.3 84.0 9.0 159.3 10.2 108.4 405.9 15.2 19.1 232.5 33.9 124.5 34.1 32.3 150.6 33.7 75.5 75.3 729.6 74.8 36.2 431.1 433.2 391.9 83.4 15.6 15.6 96.2 14.3 265.1 254.0 304.9 304.9 21.7 292.6 30.7 371.8 403.2 Model mixture: Sa, Sk, Sc For bulk average glass Model result (residual error) in normal units Residual error as % of average composition For average group 1 glass Model result (residual error) in normal units Residual error as % of average composition For average group 2 glass Model result (residual error) in normal units Residual error as % of average composition Model mixture: Sa, Sk, Sc, soil For bulk average glass Model result (residual error) in normal units Residual error as % of average composition Model mixture: Sa, Sk, Sc, KH19 For bulk average glass Model result (residual error) in normal units Residual error as % of average composition Model mixture: Sa, Sk, Sc, soil, KH19 For bulk average glass Model result (residual error) in normal units Residual error as % of average composition Table 7. Mixing model parameters and results. Table 492 K. T. Howard

Fig. 7. a) Average target rock trace element compositions normalized to group 1 glass; b) normalized to group 2 glass. rocks. The preferred interpretation is that the glass of the strewn field also show distinctly different ε-Nd values compositions reflect mixing and the isotopes are unlikely to than the glass or Eldon Group that range between +1 to −2 in have true age significance. basalts and andesites and >+5 in tholeiitic dykes (Whitford That these isotopic data support derivation of the glass et al. 1990). from Eldon Group target rocks is particularly significant Data for non-Eldon Group sedimentary rocks in the strewn when other rock formations in the strewn field are considered. field are scarce. Raheim and Compston (1977) report Rb-Sr Volcanic and igneous rocks in the strewn field can be ruled isotope data for a suite of Precambrian metasediments from out as potential targets on the basis of major and trace element near Strathgordon and the Collingwood River. These rocks are glass geochemistry that indicates a sedimentary origin. correlates of the Precambrian quartzites in the south of the Isotope data for the Mt. Read volcanics (MRV) in the center strewn field. The initial Sr ratios in these rocks (average Geochemistry of Darwin glass and target rocks from Darwin crater 493

Fig. 8. ε-Nd versus 87Sr/86Sr in glass, target rocks and crater-fill samples.

Table 8. Selected major- and trace-element abundances in Tasmanian mafic and ultramafic rocks. ppm %FeO* MgO Cr Co Ni Ni/Co Ni/Cr Cr/Co Average Tasmanian Jurassic dolerite (1) 8.8 6.6 108 30 78 2.6 0.7 3.6 Average Tasmanian west coast tholeite (2) 11.9 7.1 261 30 133 4.4 0.5 8.7 Average Tasmanian lamprophyre (3) 12.2 15.0 480 75 430 5.7 0.9 6.4 Tasmanian west coast dunites (4, 5) Average 6.2 42.0 2084 100 2547 25.5 1.2 20.8 Maximum 7.8 49.5 2630 137 3090 Minimum 5.5 33.3 1670 78 2072 85-0135 median dunite 5.8 39.4 2010 100 2470 24.7 1.2 20.1 Tasmanian west coast pyroxenites (4, 5) Average 6.2 31.8 33 4839 70 10.2 0.1 70.9 Maximum 7.4 35.6 36 5770 85 903.0 Minimum 4.3 21.0 30 3150 50 429.0 85–0161 median pyroxenite 6.1 30.0 4836 76 540 7.1 0.1 63.6 Average group 1 Darwin glass 2.2 0.9 75 9 108 12.4 1.5 8.5 Average group 2 Darwin glass 3.8 2.2 161 30 406 13.4 2.5 5.3 *All iron as FeO. Data from (1) Hergt et al. (1989); (2) Crawford and Berry (1992); (3) Baillie and Sutherland (1992); (4) Brown (1986); Keays (personal communication in 2003). 0.72678 ± 0.00207) are within the range of the Eldon Group and Explaning Cr, Co, Ni, MgO and FeO, Enrichments glass, but regressions yield divergent data arrays and distinctly in Darwin Glass older model ages (Raheim and Compston 1977). The other significant sedimentary rocks in the strewn field belong to the The single aspect of the geochemical composition of Ordovician Denison Group. These siliceous conglomerates Darwin glass that cannot be related to these suspected target contain mostly Precambrian quartzite clasts and are expected rocks from Darwin crater is the transition metal (Co,Cr,Ni), to have significantly older Rb-Sr model ages than the suspected MgO, and FeO concentrations in some group 2 glasses. Eldon Group target rocks, but data are lacking. The Denison Explaining these enrichments involves invoking an unknown Group is not associated with any suspected impact structure and mafic end-member. For comparison, analytical data for analyses suggest these conglomerates are too siliceous (SiO2 representative examples of west coast mafic and ultramafic > 97%) to be the dominant target rocks involved in the rocks have been compiled in Table 8. Average Amber Slate is formation of Darwin glass. taken to represent the Ni, Co, and Cr contributed to the glass 494 K. T. Howard

Ni/Co ratios in dunite and Darwin glass show better agreement and in the Ni/Co versus MgO plot, the mixing line between average Amber Slate and dunite passes through the group 2 glasses at up to about 10% contribution from dunite. However, these worst modelled glass compositions have excess FeO relative to average dunite at similar Ni/Co ratios. At any given FeO or MgO content, Ni/Co ratios in rocks other than the dunite (2.6–7.1) remain lower than for all group 2 glasses (average 13.4). The dunite and pyroxenite have excess Cr relative to Co when compared to all of the glasses, and trends toward higher FeO and MgO defined by the mixing lines between these rocks and Amber Slate, are divergent from the glass arrays. On the Cr/Co plots, mixing lines between Amber Slate and average basalt, dolerite and ultramafic lamprophyre define trends in the broad direction of the most enriched glasses. However, at the required FeO and MgO abundances, the dolerite, basalt, and lamprophyre also have excess Cr relative to Co. Even if allowed to contribute an unrealistically large proportion to the melt, the mixing lines show that 100% contribution from the dolerite, basalt, or lamprophyres results in excess FeO and MgO, and Cr/Co ratios would be higher than for all group 2 glass samples. Significant to note is that based on limited analyses, Tasmanian West Coast lamprophyres (Cr/Co = >15; Baillie and Sutherland 1992) appear to be more differentiated than the plotted average ultramafic lamprophyres (Cr/Co <10, Rock 1991), which can be compared to an average Cr/Co ratio of 4.7 in the most enriched glasses. This makes any contribution from lamprophyres, which are the most geologically reasonable ultramafic candidates to exist in the target stratigraphy, even less likely. As such, on the basis of excess Cr relative to Ni and Co it is not possible to mix west coast mafic or ultramafic rocks and Amber Slate, or any other analyzed suspected target rock, or multiple combinations thereof, to reproduce the Ni, Co, Cr, and MgO composition of the anomalous Fig. 9. a.) Rb-Sr isotopic evolution diagram for glass, target rocks, group 2 glass. Therefore, potential terrestrial sources of the and crater-fill samples. Data point error ellipses are 2σ. b) Sm-Nd isotopic evolution diagram for identical samples as in Fig. 9a. Data enriched transition metals in these glasses, known from the point error ellipses are 2σ. west coast of Tasmania, can largely be excluded from any further consideration. The clear implication is that these by the suspected target rocks, and linear mixing lines have anomalous enrichments may be related to the impacting been calculated between mixtures of slate and ultramafic projectile and it is perhaps significant that the transition rocks for Ni/Co, Ni/Cr and Cr/Co versus MgO (Figs. 10a–c) metals show chondritic ratios in the most anomalous glass and FeO (Figs. 10 d–f). From these plots it is clear that at any samples and this has previously been reported (Howard given MgO or FeO content, Ni/Cr ratios in all of the volcanic 2003). However, there is currently no evidence for Ir- rocks are lower than those found in the most enriched glasses. enrichment in the glass that may be expected from any The mixing line with Amber Slate indicates that even a 100% putative projectile. Meisel et al. (1990) also noted that a contribution from the most Ni-rich rock, average dunite (Ni/ chondritic contribution may partly explain the composition Cr = 1.2), could not produce Ni/Cr ratios as high as in most of the most transition-metal-enriched glasses, but they too group 2 glasses (average = 3.4) and as such the addition of could not reconcile this with the absence of any apparent Ir FeO to the average Amber Slate or mixing with highly enrichment and suggested an unknown ultra-basic unit in weathered samples like KH19 does not improve the model fit. the target stratigraphy might be a better explanation. This Geochemistry of Darwin glass and target rocks from Darwin crater 495

Fig. 10. a–f. Plots showing selected transition metal ratios versus MgO and FeO in Darwin glass, target rocks, and ultramafic rocks from western Tasmania. In each plot, linear mixing lines have been calculated between average Amber Slate and the ultramafic rocks in an attempt to model the composition of the most anomalous (high MgO, FeO, Ni, Cr, Co) glasses. The composition of the most transition-metal-enriched glass cannot be explained by a contribution from these ultramafic rocks. Data from Brown (1986), Keays (personal communication in 2003), Crawford and Berry (1992), Hergt et al. (1989), Baillie and Sutherland (1992). study has shown that contributions from known terrestrial group 2: SiO2 (76.4–84.5), Al2O3 (6.4–11.5), TiO2 (0.5– mafic and ultramafic sources in Tasmania fail to explain 0.80), FeO (1.8–5.8), MgO (1.1–4.0), K2O (1.4–2.7). the composition of the most enriched group 2 glasses. group 2 glass is also significantly enriched in Ni (117.4– 917.7), Co (19.4–56.5), and Cr (67.6–260.4) relative to CONCLUSION group 1 or bulk average Darwin glass and all of the target rocks. Data presented in this study show: • Major elements in the high SiO2 group 1 Darwin glass • Two compositional groups of glass: group 1: SiO2 samples are closest to Keel Quartzite and major elements (80.62–93.9%), Al2O3 (3.14–10.6%), TiO2 (0.2–0.76%), in the low SiO2 group 2 glasses are more similar to FeO (0.8–4.23%), MgO (0.25–2.31%), K2O (0.7–2.7%); Amber Slate. 496 K. T. Howard

• Trace elements in the glasses show affinity with both glass. Earth and Planetary Science Letters 16:228–230. Keel Quartzite (Ba, actinides, LREE) and Amber Slate Gentner W., Kirsten T., Storzer D., and Wagner G. A. 1973. K-Ar and (Sr, HREE plus Y), suggesting that the glass groups fission track dating of Darwin crater glass. Earth and Planetary Science Letters 20:204–210. represent mixtures of the shale and quartzite. Gill E. D. and Banks M. R. 1950. Silurian and Devonian stratigraphy • The glasses and suspected target rocks have concordant of the Zeehan area, Tasmania. Papers and Proceedings of the REE abundance patterns typical of upper crustal Royal Society of Tasmania. pp. 259–271. sediments. Gould C. 1866. On the position of the Gordon Limestone relative to • Sr and Nd isotope data indicate that the glasses have an other Palaeozoic formations. Papers and Proceedings of the Royal Society of Tasmania. pp. 63–66. inherited isotopic composition consistent with formation Hergt J. M., McDougall I., Banks M. R., and Green D. H. 1989. from a mixture of the suspected target rocks. Igneous rocks: Jurassic dolerite. In Geology and Mineral resources • Mixing calculations using average Eldon Group compositions of Tasmania, edited by Burrett C. F. and Martin E. L. Geological successfully model the glass composition. Such models Society of Australia Special Publication #15. pp. 375–409. result in significant errors only for Ni, Co, MgO, Cr, and Howard K. T. and Haines P. W. 2007. Geology of Darwin crater, western Tasmania, Australia. Earth and Planetary Science FeO in some anomalous group 2 glass samples. Letters 260:328–339. Therefore, the geochemical and isotopic systematics in Howard K. T. 2004. Origin of Darwin glass. Ph.D. thesis, University rocks at Darwin crater are consistent with these rocks being of Tasmania, Hobart, Australia. 362 p. the source materials melted under impact conditions to Howard K. T. 2003. Geochemical systematics in Darwin impact glass produce Darwin glass. The huge compositional heterogeneity (abstract #5079). Meteoritics & Planetary Science 38:A47. Howard K. T. and Haines P. W. 2003. Distribution and abundance of in Darwin glass indicates that this mixing of molten target Darwin impact glass (abstract # 4057). Proceedings of the Third rocks was incomplete and that the melt quenched rapidly. International Conference on Large Meteorite Impacts. Transition-metal-enriched group 2 glasses require an ultramafic Koeberl C., Reimold W. U., Blum J. D., and Chamberlain C. P. 1998. contribution. However, mixing models with dunites, Petrology and geochemistry of target rocks from the Bosumtwi pyroxenites or lamprophyres fail to produce the required glass impact structure, Ghana, and comparison with Ivory Coast tektites. Geochimica et Cosmochimica Acta 62:2179–2196. compositions and these rock types are not present in the target Loh C. H., Howard K. T., Chung S. L., and Meffre S. 2002. Laser rock stratigraphy. The implication is that these transition fusion 40Ar/39Ar ages of Darwin Impact glass. Meteoritics & metal-enriched glasses are preserving the geochemical Planetary Science 37:1555–1562. signature of the impacting projectile. Meisel T., Koeberl C., and Ford R. J. 1990. Geochemistry of Darwin impact glass and target rocks. Geochimica et Cosmochimica Acta 54:1463–1474. Acknowledgments–Peter Haines, Ron Berry, Clive Burrett, Raheim A. and Compston W. 1977. Correlations between metamorphic and Marc Norman provided valued insights. Parts of this events and Rb-Sr ages in metasediments and eclogite from research were funded by the Barringer Family Fund for western Tasmania. Lithos 10:271–289. Meteorite Impact Research Award received by the author in 2003. Rock N. M. S. 1991. Lamprophyres. Glasgow, UK: Blackie. 285 p. Two anonymous reviewers are thanked for their constructive Schaaf P. and Müller-Sohnius D. M. 2002. Strontium and neodymium isotopic study of Libyan Desert Glass: Inherited comments. The Tasmanian Aboriginal people are the traditional pan-African age signatures and new evidence for target material. owners of Darwin glass. Meteoritics & Planetary Science 37:565–576. Shaw H. F. and Wasserburg G. J. 1982. Age and provenance of the Editorial Handling—Dr. John Spray target materials for tektites and possible impactites as inferred from Sm-Nd and Rb-Sr systematics. Earth and Planetary Science Letters 60:155–177. REFERENCES Sun S.-S. and McDonough W. F. 1989. Chemical and isotopic systematics of oceanic basalts: Implications for mantle Baillie P. W. and Sutherland F. L. 1992. Devonian lamprophyres from composition and processes. In Magmatism in the ocean basins, Mt. Lyell, western Tasmania. Papers and Proceedings of the edited by Saunders A. D. and Norry M. J. London: Geological Royal Society of Tasmania 126:19–22. Society. pp. 313–345. Brown A. V. 1986. Geology of the Dundas-Mt. Lindsay-Mt. Whitehead J., Papanastassiou D. A., Spray J. G., Grieve R. A. F., and Youngback region. Tasmania Department of Mines. Geolocial Wasserburg G. J. 2000. Late Eocene impact ejecta: Geochemical Survey Bulletin No. 62. 218 p. and isotopic connections with the Popigai impact structure. Earth Corbett K. D. and Brown A. V. 1975. Geological atlas 1:250,000 and Planetary Science Letters 181:473–487. series, Queenstown. Geological Survey of Tasmania, Department Whitford D. J., Crawford A. J., Korsch M. J., and Craven S. J. of Mines. 1990. Strontium and neodymium isotopic studies of the Corbett K. D., Pemberton J., and Vicary M. J. 1993. Geology of the Mount Read volcanics, Tasmania. Proceedings of Tenth Mt. Jukes-Mt. Darwin area, 1:25,000 Map 13. Geological Australian Geologic Convention. Geologic Society of Survey of Tasmania, Department of Mines. Australia. pp. 215–216. Crawford A. J. and Berry R. F. 1992. Tectonic implications of late Williams E. A. 1989. Summary and synthesis. In Geology and Proterozoic-early Palaeozoic igneous rock associations in mineral resources of Tasmania, edited by Burrett C. F. and Martin western Tasmania. Tectonophysics 214:37–56. E. L. Geological Society of Australia Special Publication #15. Ford R. J. 1972. A possible impact crater associated with Darwin pp. 468–515.